Hi Rob,

The data I have does not have a PK. It is basically a dump for a year from many locations and used for analysis on the avaliable fields.

While the missing list is expected to have hardly anything, there can be a case where the reference data may be missing completely making the Sort component (and may be the server as well) go for a toss.

Moreover, if even one of the values from a row (for CourseID) is missing, the entire record should be discarded. At the same time, other fields (non CourseID fields) should still be validated and all the failed validations should be consolidated in a comments field for the user.

Lastly, the reference dataset I use is filtered and has ~15K rows in one scenario and ~50K rows in another. The underlying table may have much more data.